Beyond the Prompt Box: How Hapax Is Bringing Proactive AI to Credit Unions

By John San Filippo

As generative artificial intelligence continues to make inroads across financial services, many credit unions are wrestling with the same challenge: managing an assortment of disconnected point solutions or deploying general chat tools without a clear strategic roadmap. Austin-based Hapax is positioning itself to solve that problem by pivoting from static knowledge bases to an active, unified operational intelligence platform.

Finopotamus recently sat down with Hapax CRO Aaron Kwan to learn more about his company’s pivot. He outlined how the fintech revamped its architecture to deliver what he described as “proactive artificial intelligence.” Rather than requiring staff to come up with complex prompts, the system actively identifies inefficiencies and builds automated agents and workflows based on those findings.

Moving from Reactive Prompts to Proactive Agents

Aaron Kwan

Created in 2024 by Hank Seale, founder of Q2 Software, Hapax initially focused on ingesting credit union policies and procedures to function as an internal knowledge management tool. However, the company recognized that the market for standard smart knowledge bases was becoming increasingly crowded. In response, Hapax shifted gears toward proactive operations.

“Rather than sitting back and waiting for prompts, it builds things for you,” Kwan explained. “The point of Hapax is that it learns about your business.” He added the software creates an individualized experience for each user.

The onboarding process begins with a dynamic 12-to-15-minute interview conducted via text or conversational voice. During this setup, the platform learns the details of a user’s role, team structure, and workflows. From there, it integrates into daily workplace applications – such as email, Slack, calendars, Google Drive, SharePoint, and CRMs like HubSpot – using MCP connectors.

Once connected, Hapax observes daily operations to identify repetitive tasks, evaluate workflows, and recommend specific automations. The platform includes a catalog of roughly 150 pre-packaged workflows designed for financial institutions, while also allowing users and institutions to create tailored agents.

Transforming the Back Office and Data Intelligence

While many market offerings focus strictly on member-facing chatbots, Hapax focuses squarely on internal operations and back-office efficiency, Kwan noted. “We're looking for opportunities to make the credit union more efficient and eliminate remedial tasks,” he continued. “We're not looking to replace people.”

One major use case involves high-volume document and data processing. For example, Kwan cited an indirect lending workflow where loan applications from some 90 auto dealerships are routed directly to an email address. Rather than having multiple employees manually audit paper documents line by line for compliance and accuracy, the platform ingests the files and verifies compliance within minutes.

Beyond task automation, Hapax acts as an operational hub connecting to core systems and data lakes, such as Snowflake, or ingesting rich transaction files like Q2's Eve extracts. This enables credit union teams to run complex behavioral queries using natural language.

Kwan highlighted a real-world case where an institution used the platform to analyze behavioral patterns among departing accountholders under age 30, identify current members exhibiting those same behaviors, and route those accounts into targeted retention workflows.

Built for Regulatory Compliance

Operating within a highly regulated environment means security and auditability are non-negotiable. Hapax maintains SOC 2 compliance and records an audit trail for every action taken within the platform.

“Everything that takes place inside of our platform, there's an audit trail,” Kwan said. “You can see who did what, the reasoning, what the prompt was, or if there were agents that were used, etc.”

To optimize accuracy and control operational costs, Hapax does not rely on a single foundation model. Instead, it routes requests across 75 different large language models using least-cost routing, matching specific tasks to the most suitable engine while maintaining failover redundancy.

With several billion-dollar credit unions either signed up or already deployed, Hapax is determined to demonstrate how comprehensive internal AI architectures can replace fragmented point solutions with measurable operational gains.

While any number of solutions allow a credit union to differentiate their member experience, Kwan said that Hapax empowers credit unions to differentiate themselves operationally by creating a highly customized back-office environment. “This is exactly what Hapax is for,” he told Finopotamus.